LLaVA-Tri-PathVQA / trainer_state.json
yunfeixie's picture
Add files using upload-large-folder tool
eb8817c verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 462,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.4285714285714286e-06,
"loss": 9.0491,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 2.8571428571428573e-06,
"loss": 9.4873,
"step": 2
},
{
"epoch": 0.02,
"learning_rate": 4.2857142857142855e-06,
"loss": 8.9786,
"step": 3
},
{
"epoch": 0.03,
"learning_rate": 5.7142857142857145e-06,
"loss": 7.06,
"step": 4
},
{
"epoch": 0.03,
"learning_rate": 7.1428571428571436e-06,
"loss": 4.2462,
"step": 5
},
{
"epoch": 0.04,
"learning_rate": 8.571428571428571e-06,
"loss": 3.9334,
"step": 6
},
{
"epoch": 0.05,
"learning_rate": 1e-05,
"loss": 2.0373,
"step": 7
},
{
"epoch": 0.05,
"learning_rate": 1.1428571428571429e-05,
"loss": 2.0989,
"step": 8
},
{
"epoch": 0.06,
"learning_rate": 1.2857142857142859e-05,
"loss": 1.8871,
"step": 9
},
{
"epoch": 0.06,
"learning_rate": 1.4285714285714287e-05,
"loss": 2.0296,
"step": 10
},
{
"epoch": 0.07,
"learning_rate": 1.5714285714285715e-05,
"loss": 1.3407,
"step": 11
},
{
"epoch": 0.08,
"learning_rate": 1.7142857142857142e-05,
"loss": 1.393,
"step": 12
},
{
"epoch": 0.08,
"learning_rate": 1.8571428571428575e-05,
"loss": 1.266,
"step": 13
},
{
"epoch": 0.09,
"learning_rate": 2e-05,
"loss": 1.4216,
"step": 14
},
{
"epoch": 0.1,
"learning_rate": 1.9999754126376247e-05,
"loss": 1.4259,
"step": 15
},
{
"epoch": 0.1,
"learning_rate": 1.9999016517595752e-05,
"loss": 1.4195,
"step": 16
},
{
"epoch": 0.11,
"learning_rate": 1.9997787209930222e-05,
"loss": 1.6327,
"step": 17
},
{
"epoch": 0.12,
"learning_rate": 1.9996066263830533e-05,
"loss": 1.1754,
"step": 18
},
{
"epoch": 0.12,
"learning_rate": 1.9993853763923724e-05,
"loss": 1.4505,
"step": 19
},
{
"epoch": 0.13,
"learning_rate": 1.999114981900887e-05,
"loss": 1.2528,
"step": 20
},
{
"epoch": 0.14,
"learning_rate": 1.9987954562051724e-05,
"loss": 1.3553,
"step": 21
},
{
"epoch": 0.14,
"learning_rate": 1.998426815017817e-05,
"loss": 1.1403,
"step": 22
},
{
"epoch": 0.15,
"learning_rate": 1.9980090764666486e-05,
"loss": 1.5534,
"step": 23
},
{
"epoch": 0.16,
"learning_rate": 1.9975422610938463e-05,
"loss": 1.1475,
"step": 24
},
{
"epoch": 0.16,
"learning_rate": 1.9970263918549274e-05,
"loss": 1.3228,
"step": 25
},
{
"epoch": 0.17,
"learning_rate": 1.9964614941176194e-05,
"loss": 1.4057,
"step": 26
},
{
"epoch": 0.18,
"learning_rate": 1.9958475956606133e-05,
"loss": 1.6701,
"step": 27
},
{
"epoch": 0.18,
"learning_rate": 1.995184726672197e-05,
"loss": 1.4343,
"step": 28
},
{
"epoch": 0.19,
"learning_rate": 1.9944729197487702e-05,
"loss": 1.4512,
"step": 29
},
{
"epoch": 0.19,
"learning_rate": 1.9937122098932428e-05,
"loss": 1.2664,
"step": 30
},
{
"epoch": 0.2,
"learning_rate": 1.992902634513312e-05,
"loss": 1.0385,
"step": 31
},
{
"epoch": 0.21,
"learning_rate": 1.9920442334196248e-05,
"loss": 1.2659,
"step": 32
},
{
"epoch": 0.21,
"learning_rate": 1.9911370488238185e-05,
"loss": 1.5775,
"step": 33
},
{
"epoch": 0.22,
"learning_rate": 1.9901811253364458e-05,
"loss": 1.4111,
"step": 34
},
{
"epoch": 0.23,
"learning_rate": 1.989176509964781e-05,
"loss": 1.332,
"step": 35
},
{
"epoch": 0.23,
"learning_rate": 1.988123252110509e-05,
"loss": 1.2439,
"step": 36
},
{
"epoch": 0.24,
"learning_rate": 1.9870214035672945e-05,
"loss": 1.4402,
"step": 37
},
{
"epoch": 0.25,
"learning_rate": 1.985871018518236e-05,
"loss": 1.155,
"step": 38
},
{
"epoch": 0.25,
"learning_rate": 1.984672153533202e-05,
"loss": 1.3743,
"step": 39
},
{
"epoch": 0.26,
"learning_rate": 1.9834248675660484e-05,
"loss": 1.1744,
"step": 40
},
{
"epoch": 0.27,
"learning_rate": 1.982129221951719e-05,
"loss": 0.7612,
"step": 41
},
{
"epoch": 0.27,
"learning_rate": 1.9807852804032306e-05,
"loss": 1.1555,
"step": 42
},
{
"epoch": 0.28,
"learning_rate": 1.9793931090085385e-05,
"loss": 1.2492,
"step": 43
},
{
"epoch": 0.29,
"learning_rate": 1.9779527762272877e-05,
"loss": 0.984,
"step": 44
},
{
"epoch": 0.29,
"learning_rate": 1.976464352887447e-05,
"loss": 1.0539,
"step": 45
},
{
"epoch": 0.3,
"learning_rate": 1.9749279121818235e-05,
"loss": 1.2939,
"step": 46
},
{
"epoch": 0.31,
"learning_rate": 1.973343529664467e-05,
"loss": 1.3436,
"step": 47
},
{
"epoch": 0.31,
"learning_rate": 1.971711283246951e-05,
"loss": 1.1927,
"step": 48
},
{
"epoch": 0.32,
"learning_rate": 1.9700312531945444e-05,
"loss": 1.3123,
"step": 49
},
{
"epoch": 0.32,
"learning_rate": 1.9683035221222617e-05,
"loss": 1.5792,
"step": 50
},
{
"epoch": 0.33,
"learning_rate": 1.9665281749908034e-05,
"loss": 1.0083,
"step": 51
},
{
"epoch": 0.34,
"learning_rate": 1.964705299102376e-05,
"loss": 1.2839,
"step": 52
},
{
"epoch": 0.34,
"learning_rate": 1.9628349840963997e-05,
"loss": 1.1535,
"step": 53
},
{
"epoch": 0.35,
"learning_rate": 1.9609173219450998e-05,
"loss": 1.4899,
"step": 54
},
{
"epoch": 0.36,
"learning_rate": 1.958952406948985e-05,
"loss": 1.2932,
"step": 55
},
{
"epoch": 0.36,
"learning_rate": 1.956940335732209e-05,
"loss": 1.4064,
"step": 56
},
{
"epoch": 0.37,
"learning_rate": 1.9548812072378208e-05,
"loss": 1.0525,
"step": 57
},
{
"epoch": 0.38,
"learning_rate": 1.9527751227228964e-05,
"loss": 1.2386,
"step": 58
},
{
"epoch": 0.38,
"learning_rate": 1.950622185753563e-05,
"loss": 1.3053,
"step": 59
},
{
"epoch": 0.39,
"learning_rate": 1.9484225021999032e-05,
"loss": 1.1901,
"step": 60
},
{
"epoch": 0.4,
"learning_rate": 1.9461761802307494e-05,
"loss": 1.2362,
"step": 61
},
{
"epoch": 0.4,
"learning_rate": 1.9438833303083677e-05,
"loss": 1.306,
"step": 62
},
{
"epoch": 0.41,
"learning_rate": 1.941544065183021e-05,
"loss": 1.114,
"step": 63
},
{
"epoch": 0.42,
"learning_rate": 1.939158499887428e-05,
"loss": 0.9631,
"step": 64
},
{
"epoch": 0.42,
"learning_rate": 1.9367267517311057e-05,
"loss": 1.0238,
"step": 65
},
{
"epoch": 0.43,
"learning_rate": 1.9342489402945997e-05,
"loss": 1.0693,
"step": 66
},
{
"epoch": 0.44,
"learning_rate": 1.9317251874236066e-05,
"loss": 1.3379,
"step": 67
},
{
"epoch": 0.44,
"learning_rate": 1.9291556172229784e-05,
"loss": 1.2689,
"step": 68
},
{
"epoch": 0.45,
"learning_rate": 1.9265403560506223e-05,
"loss": 1.1508,
"step": 69
},
{
"epoch": 0.45,
"learning_rate": 1.9238795325112867e-05,
"loss": 1.1662,
"step": 70
},
{
"epoch": 0.46,
"learning_rate": 1.9211732774502372e-05,
"loss": 1.3371,
"step": 71
},
{
"epoch": 0.47,
"learning_rate": 1.9184217239468213e-05,
"loss": 0.9793,
"step": 72
},
{
"epoch": 0.47,
"learning_rate": 1.915625007307925e-05,
"loss": 1.0375,
"step": 73
},
{
"epoch": 0.48,
"learning_rate": 1.912783265061319e-05,
"loss": 1.2865,
"step": 74
},
{
"epoch": 0.49,
"learning_rate": 1.9098966369488967e-05,
"loss": 0.937,
"step": 75
},
{
"epoch": 0.49,
"learning_rate": 1.9069652649198004e-05,
"loss": 1.2003,
"step": 76
},
{
"epoch": 0.5,
"learning_rate": 1.9039892931234434e-05,
"loss": 1.1301,
"step": 77
},
{
"epoch": 0.51,
"learning_rate": 1.900968867902419e-05,
"loss": 1.0041,
"step": 78
},
{
"epoch": 0.51,
"learning_rate": 1.8979041377853068e-05,
"loss": 1.128,
"step": 79
},
{
"epoch": 0.52,
"learning_rate": 1.8947952534793663e-05,
"loss": 1.0773,
"step": 80
},
{
"epoch": 0.53,
"learning_rate": 1.891642367863127e-05,
"loss": 0.8535,
"step": 81
},
{
"epoch": 0.53,
"learning_rate": 1.8884456359788725e-05,
"loss": 1.0934,
"step": 82
},
{
"epoch": 0.54,
"learning_rate": 1.8852052150250123e-05,
"loss": 1.2063,
"step": 83
},
{
"epoch": 0.55,
"learning_rate": 1.881921264348355e-05,
"loss": 0.7568,
"step": 84
},
{
"epoch": 0.55,
"learning_rate": 1.878593945436272e-05,
"loss": 1.0244,
"step": 85
},
{
"epoch": 0.56,
"learning_rate": 1.8752234219087538e-05,
"loss": 1.2583,
"step": 86
},
{
"epoch": 0.56,
"learning_rate": 1.871809859510368e-05,
"loss": 1.0935,
"step": 87
},
{
"epoch": 0.57,
"learning_rate": 1.8683534261021058e-05,
"loss": 1.1058,
"step": 88
},
{
"epoch": 0.58,
"learning_rate": 1.8648542916531283e-05,
"loss": 0.8688,
"step": 89
},
{
"epoch": 0.58,
"learning_rate": 1.8613126282324092e-05,
"loss": 0.8368,
"step": 90
},
{
"epoch": 0.59,
"learning_rate": 1.8577286100002723e-05,
"loss": 1.2209,
"step": 91
},
{
"epoch": 0.6,
"learning_rate": 1.8541024131998277e-05,
"loss": 1.2839,
"step": 92
},
{
"epoch": 0.6,
"learning_rate": 1.850434216148305e-05,
"loss": 0.9052,
"step": 93
},
{
"epoch": 0.61,
"learning_rate": 1.8467241992282842e-05,
"loss": 1.3185,
"step": 94
},
{
"epoch": 0.62,
"learning_rate": 1.8429725448788267e-05,
"loss": 1.1049,
"step": 95
},
{
"epoch": 0.62,
"learning_rate": 1.8391794375865025e-05,
"loss": 1.2833,
"step": 96
},
{
"epoch": 0.63,
"learning_rate": 1.8353450638763178e-05,
"loss": 1.2236,
"step": 97
},
{
"epoch": 0.64,
"learning_rate": 1.8314696123025456e-05,
"loss": 0.9664,
"step": 98
},
{
"epoch": 0.64,
"learning_rate": 1.827553273439449e-05,
"loss": 0.9863,
"step": 99
},
{
"epoch": 0.65,
"learning_rate": 1.823596239871915e-05,
"loss": 1.1943,
"step": 100
},
{
"epoch": 0.66,
"learning_rate": 1.819598706185979e-05,
"loss": 1.1481,
"step": 101
},
{
"epoch": 0.66,
"learning_rate": 1.8155608689592604e-05,
"loss": 1.0814,
"step": 102
},
{
"epoch": 0.67,
"learning_rate": 1.811482926751293e-05,
"loss": 0.9433,
"step": 103
},
{
"epoch": 0.68,
"learning_rate": 1.8073650800937627e-05,
"loss": 1.0319,
"step": 104
},
{
"epoch": 0.68,
"learning_rate": 1.803207531480645e-05,
"loss": 1.0255,
"step": 105
},
{
"epoch": 0.69,
"learning_rate": 1.7990104853582494e-05,
"loss": 1.1507,
"step": 106
},
{
"epoch": 0.69,
"learning_rate": 1.7947741481151628e-05,
"loss": 0.8641,
"step": 107
},
{
"epoch": 0.7,
"learning_rate": 1.7904987280721037e-05,
"loss": 0.8619,
"step": 108
},
{
"epoch": 0.71,
"learning_rate": 1.7861844354716757e-05,
"loss": 0.9446,
"step": 109
},
{
"epoch": 0.71,
"learning_rate": 1.78183148246803e-05,
"loss": 1.0374,
"step": 110
},
{
"epoch": 0.72,
"learning_rate": 1.777440083116432e-05,
"loss": 0.9664,
"step": 111
},
{
"epoch": 0.73,
"learning_rate": 1.773010453362737e-05,
"loss": 0.8105,
"step": 112
},
{
"epoch": 0.73,
"learning_rate": 1.7685428110327683e-05,
"loss": 1.1674,
"step": 113
},
{
"epoch": 0.74,
"learning_rate": 1.7640373758216075e-05,
"loss": 1.1545,
"step": 114
},
{
"epoch": 0.75,
"learning_rate": 1.7594943692827913e-05,
"loss": 0.7447,
"step": 115
},
{
"epoch": 0.75,
"learning_rate": 1.754914014817416e-05,
"loss": 1.056,
"step": 116
},
{
"epoch": 0.76,
"learning_rate": 1.7502965376631515e-05,
"loss": 0.8895,
"step": 117
},
{
"epoch": 0.77,
"learning_rate": 1.7456421648831658e-05,
"loss": 1.2917,
"step": 118
},
{
"epoch": 0.77,
"learning_rate": 1.7409511253549592e-05,
"loss": 1.1788,
"step": 119
},
{
"epoch": 0.78,
"learning_rate": 1.7362236497591097e-05,
"loss": 1.0486,
"step": 120
},
{
"epoch": 0.79,
"learning_rate": 1.731459970567928e-05,
"loss": 1.0283,
"step": 121
},
{
"epoch": 0.79,
"learning_rate": 1.7266603220340273e-05,
"loss": 1.0439,
"step": 122
},
{
"epoch": 0.8,
"learning_rate": 1.7218249401788033e-05,
"loss": 1.1714,
"step": 123
},
{
"epoch": 0.81,
"learning_rate": 1.7169540627808276e-05,
"loss": 1.1786,
"step": 124
},
{
"epoch": 0.81,
"learning_rate": 1.7120479293641558e-05,
"loss": 1.0071,
"step": 125
},
{
"epoch": 0.82,
"learning_rate": 1.7071067811865477e-05,
"loss": 1.1251,
"step": 126
},
{
"epoch": 0.82,
"learning_rate": 1.7021308612276056e-05,
"loss": 1.0195,
"step": 127
},
{
"epoch": 0.83,
"learning_rate": 1.6971204141768235e-05,
"loss": 1.1006,
"step": 128
},
{
"epoch": 0.84,
"learning_rate": 1.6920756864215558e-05,
"loss": 1.0999,
"step": 129
},
{
"epoch": 0.84,
"learning_rate": 1.686996926034902e-05,
"loss": 1.1979,
"step": 130
},
{
"epoch": 0.85,
"learning_rate": 1.6818843827635052e-05,
"loss": 0.7987,
"step": 131
},
{
"epoch": 0.86,
"learning_rate": 1.6767383080152744e-05,
"loss": 0.6544,
"step": 132
},
{
"epoch": 0.86,
"learning_rate": 1.6715589548470187e-05,
"loss": 1.0395,
"step": 133
},
{
"epoch": 0.87,
"learning_rate": 1.6663465779520042e-05,
"loss": 0.9087,
"step": 134
},
{
"epoch": 0.88,
"learning_rate": 1.6611014336474303e-05,
"loss": 1.0207,
"step": 135
},
{
"epoch": 0.88,
"learning_rate": 1.6558237798618243e-05,
"loss": 0.9879,
"step": 136
},
{
"epoch": 0.89,
"learning_rate": 1.6505138761223586e-05,
"loss": 0.9363,
"step": 137
},
{
"epoch": 0.9,
"learning_rate": 1.645171983542088e-05,
"loss": 0.8757,
"step": 138
},
{
"epoch": 0.9,
"learning_rate": 1.6397983648071093e-05,
"loss": 0.955,
"step": 139
},
{
"epoch": 0.91,
"learning_rate": 1.6343932841636455e-05,
"loss": 0.9536,
"step": 140
},
{
"epoch": 0.92,
"learning_rate": 1.6289570074050492e-05,
"loss": 1.1107,
"step": 141
},
{
"epoch": 0.92,
"learning_rate": 1.6234898018587336e-05,
"loss": 0.9642,
"step": 142
},
{
"epoch": 0.93,
"learning_rate": 1.617991936373027e-05,
"loss": 0.8894,
"step": 143
},
{
"epoch": 0.94,
"learning_rate": 1.6124636813039502e-05,
"loss": 0.9625,
"step": 144
},
{
"epoch": 0.94,
"learning_rate": 1.6069053085019258e-05,
"loss": 0.9687,
"step": 145
},
{
"epoch": 0.95,
"learning_rate": 1.601317091298406e-05,
"loss": 0.882,
"step": 146
},
{
"epoch": 0.95,
"learning_rate": 1.5956993044924334e-05,
"loss": 0.781,
"step": 147
},
{
"epoch": 0.96,
"learning_rate": 1.5900522243371283e-05,
"loss": 0.6522,
"step": 148
},
{
"epoch": 0.97,
"learning_rate": 1.5843761285261027e-05,
"loss": 0.9439,
"step": 149
},
{
"epoch": 0.97,
"learning_rate": 1.578671296179806e-05,
"loss": 0.9486,
"step": 150
},
{
"epoch": 0.98,
"learning_rate": 1.5729380078317982e-05,
"loss": 1.0371,
"step": 151
},
{
"epoch": 0.99,
"learning_rate": 1.5671765454149558e-05,
"loss": 1.3166,
"step": 152
},
{
"epoch": 0.99,
"learning_rate": 1.5613871922476082e-05,
"loss": 1.0802,
"step": 153
},
{
"epoch": 1.0,
"learning_rate": 1.5555702330196024e-05,
"loss": 1.0281,
"step": 154
},
{
"epoch": 1.01,
"learning_rate": 1.5497259537783084e-05,
"loss": 0.3437,
"step": 155
},
{
"epoch": 1.01,
"learning_rate": 1.543854641914549e-05,
"loss": 0.6031,
"step": 156
},
{
"epoch": 1.02,
"learning_rate": 1.537956586148469e-05,
"loss": 0.582,
"step": 157
},
{
"epoch": 1.03,
"learning_rate": 1.5320320765153367e-05,
"loss": 0.4816,
"step": 158
},
{
"epoch": 1.03,
"learning_rate": 1.5260814043512838e-05,
"loss": 0.4704,
"step": 159
},
{
"epoch": 1.04,
"learning_rate": 1.5201048622789747e-05,
"loss": 0.3908,
"step": 160
},
{
"epoch": 1.05,
"learning_rate": 1.5141027441932217e-05,
"loss": 0.3807,
"step": 161
},
{
"epoch": 1.05,
"learning_rate": 1.5080753452465296e-05,
"loss": 0.3841,
"step": 162
},
{
"epoch": 1.06,
"learning_rate": 1.502022961834582e-05,
"loss": 0.4762,
"step": 163
},
{
"epoch": 1.06,
"learning_rate": 1.4959458915816681e-05,
"loss": 0.5369,
"step": 164
},
{
"epoch": 1.07,
"learning_rate": 1.4898444333260436e-05,
"loss": 0.4141,
"step": 165
},
{
"epoch": 1.08,
"learning_rate": 1.4837188871052399e-05,
"loss": 0.4642,
"step": 166
},
{
"epoch": 1.08,
"learning_rate": 1.4775695541413063e-05,
"loss": 0.3297,
"step": 167
},
{
"epoch": 1.09,
"learning_rate": 1.4713967368259981e-05,
"loss": 0.3809,
"step": 168
},
{
"epoch": 1.1,
"learning_rate": 1.4652007387059077e-05,
"loss": 0.4052,
"step": 169
},
{
"epoch": 1.1,
"learning_rate": 1.4589818644675378e-05,
"loss": 0.4155,
"step": 170
},
{
"epoch": 1.11,
"learning_rate": 1.4527404199223173e-05,
"loss": 0.4513,
"step": 171
},
{
"epoch": 1.12,
"learning_rate": 1.446476711991563e-05,
"loss": 0.3982,
"step": 172
},
{
"epoch": 1.12,
"learning_rate": 1.4401910486913892e-05,
"loss": 0.4712,
"step": 173
},
{
"epoch": 1.13,
"learning_rate": 1.4338837391175582e-05,
"loss": 0.3968,
"step": 174
},
{
"epoch": 1.14,
"learning_rate": 1.4275550934302822e-05,
"loss": 0.5116,
"step": 175
},
{
"epoch": 1.14,
"learning_rate": 1.4212054228389712e-05,
"loss": 0.5018,
"step": 176
},
{
"epoch": 1.15,
"learning_rate": 1.4148350395869279e-05,
"loss": 0.4874,
"step": 177
},
{
"epoch": 1.16,
"learning_rate": 1.4084442569359964e-05,
"loss": 0.4139,
"step": 178
},
{
"epoch": 1.16,
"learning_rate": 1.4020333891511536e-05,
"loss": 0.4595,
"step": 179
},
{
"epoch": 1.17,
"learning_rate": 1.395602751485059e-05,
"loss": 0.3834,
"step": 180
},
{
"epoch": 1.18,
"learning_rate": 1.3891526601625492e-05,
"loss": 0.5345,
"step": 181
},
{
"epoch": 1.18,
"learning_rate": 1.3826834323650899e-05,
"loss": 0.3391,
"step": 182
},
{
"epoch": 1.19,
"learning_rate": 1.3761953862151773e-05,
"loss": 0.3089,
"step": 183
},
{
"epoch": 1.19,
"learning_rate": 1.3696888407606952e-05,
"loss": 0.4527,
"step": 184
},
{
"epoch": 1.2,
"learning_rate": 1.3631641159592253e-05,
"loss": 0.3471,
"step": 185
},
{
"epoch": 1.21,
"learning_rate": 1.3566215326623131e-05,
"loss": 0.4585,
"step": 186
},
{
"epoch": 1.21,
"learning_rate": 1.3500614125996924e-05,
"loss": 0.4888,
"step": 187
},
{
"epoch": 1.22,
"learning_rate": 1.3434840783634611e-05,
"loss": 0.4484,
"step": 188
},
{
"epoch": 1.23,
"learning_rate": 1.3368898533922202e-05,
"loss": 0.5122,
"step": 189
},
{
"epoch": 1.23,
"learning_rate": 1.3302790619551673e-05,
"loss": 0.4414,
"step": 190
},
{
"epoch": 1.24,
"learning_rate": 1.3236520291361516e-05,
"loss": 0.4887,
"step": 191
},
{
"epoch": 1.25,
"learning_rate": 1.3170090808176883e-05,
"loss": 0.4562,
"step": 192
},
{
"epoch": 1.25,
"learning_rate": 1.310350543664932e-05,
"loss": 0.3805,
"step": 193
},
{
"epoch": 1.26,
"learning_rate": 1.3036767451096148e-05,
"loss": 0.3698,
"step": 194
},
{
"epoch": 1.27,
"learning_rate": 1.2969880133339437e-05,
"loss": 0.4023,
"step": 195
},
{
"epoch": 1.27,
"learning_rate": 1.2902846772544625e-05,
"loss": 0.4144,
"step": 196
},
{
"epoch": 1.28,
"learning_rate": 1.2835670665058779e-05,
"loss": 0.3273,
"step": 197
},
{
"epoch": 1.29,
"learning_rate": 1.2768355114248493e-05,
"loss": 0.5584,
"step": 198
},
{
"epoch": 1.29,
"learning_rate": 1.2700903430337456e-05,
"loss": 0.4242,
"step": 199
},
{
"epoch": 1.3,
"learning_rate": 1.2633318930243647e-05,
"loss": 0.4097,
"step": 200
},
{
"epoch": 1.31,
"learning_rate": 1.2565604937416267e-05,
"loss": 0.4673,
"step": 201
},
{
"epoch": 1.31,
"learning_rate": 1.249776478167227e-05,
"loss": 0.3573,
"step": 202
},
{
"epoch": 1.32,
"learning_rate": 1.242980179903264e-05,
"loss": 0.3535,
"step": 203
},
{
"epoch": 1.32,
"learning_rate": 1.2361719331558346e-05,
"loss": 0.3292,
"step": 204
},
{
"epoch": 1.33,
"learning_rate": 1.229352072718598e-05,
"loss": 0.35,
"step": 205
},
{
"epoch": 1.34,
"learning_rate": 1.2225209339563144e-05,
"loss": 0.3676,
"step": 206
},
{
"epoch": 1.34,
"learning_rate": 1.2156788527883524e-05,
"loss": 0.4234,
"step": 207
},
{
"epoch": 1.35,
"learning_rate": 1.20882616567217e-05,
"loss": 0.3414,
"step": 208
},
{
"epoch": 1.36,
"learning_rate": 1.2019632095867697e-05,
"loss": 0.475,
"step": 209
},
{
"epoch": 1.36,
"learning_rate": 1.1950903220161286e-05,
"loss": 0.324,
"step": 210
},
{
"epoch": 1.37,
"learning_rate": 1.1882078409326003e-05,
"loss": 0.4944,
"step": 211
},
{
"epoch": 1.38,
"learning_rate": 1.1813161047802986e-05,
"loss": 0.5163,
"step": 212
},
{
"epoch": 1.38,
"learning_rate": 1.174415452458451e-05,
"loss": 0.3491,
"step": 213
},
{
"epoch": 1.39,
"learning_rate": 1.1675062233047365e-05,
"loss": 0.4207,
"step": 214
},
{
"epoch": 1.4,
"learning_rate": 1.1605887570785972e-05,
"loss": 0.4161,
"step": 215
},
{
"epoch": 1.4,
"learning_rate": 1.1536633939445302e-05,
"loss": 0.5664,
"step": 216
},
{
"epoch": 1.41,
"learning_rate": 1.1467304744553618e-05,
"loss": 0.4413,
"step": 217
},
{
"epoch": 1.42,
"learning_rate": 1.1397903395354996e-05,
"loss": 0.3715,
"step": 218
},
{
"epoch": 1.42,
"learning_rate": 1.132843330464168e-05,
"loss": 0.4286,
"step": 219
},
{
"epoch": 1.43,
"learning_rate": 1.1258897888586256e-05,
"loss": 0.413,
"step": 220
},
{
"epoch": 1.44,
"learning_rate": 1.118930056657367e-05,
"loss": 0.2996,
"step": 221
},
{
"epoch": 1.44,
"learning_rate": 1.1119644761033079e-05,
"loss": 0.4393,
"step": 222
},
{
"epoch": 1.45,
"learning_rate": 1.1049933897269547e-05,
"loss": 0.4494,
"step": 223
},
{
"epoch": 1.45,
"learning_rate": 1.098017140329561e-05,
"loss": 0.3295,
"step": 224
},
{
"epoch": 1.46,
"learning_rate": 1.0910360709662701e-05,
"loss": 0.5429,
"step": 225
},
{
"epoch": 1.47,
"learning_rate": 1.0840505249292477e-05,
"loss": 0.4348,
"step": 226
},
{
"epoch": 1.47,
"learning_rate": 1.0770608457307965e-05,
"loss": 0.4108,
"step": 227
},
{
"epoch": 1.48,
"learning_rate": 1.0700673770864673e-05,
"loss": 0.4371,
"step": 228
},
{
"epoch": 1.49,
"learning_rate": 1.0630704628981561e-05,
"loss": 0.4465,
"step": 229
},
{
"epoch": 1.49,
"learning_rate": 1.0560704472371919e-05,
"loss": 0.4459,
"step": 230
},
{
"epoch": 1.5,
"learning_rate": 1.0490676743274181e-05,
"loss": 0.4298,
"step": 231
},
{
"epoch": 1.51,
"learning_rate": 1.0420624885282653e-05,
"loss": 0.4779,
"step": 232
},
{
"epoch": 1.51,
"learning_rate": 1.0350552343178164e-05,
"loss": 0.4978,
"step": 233
},
{
"epoch": 1.52,
"learning_rate": 1.028046256275869e-05,
"loss": 0.4915,
"step": 234
},
{
"epoch": 1.53,
"learning_rate": 1.0210358990669889e-05,
"loss": 0.4305,
"step": 235
},
{
"epoch": 1.53,
"learning_rate": 1.0140245074235624e-05,
"loss": 0.4079,
"step": 236
},
{
"epoch": 1.54,
"learning_rate": 1.0070124261288437e-05,
"loss": 0.4446,
"step": 237
},
{
"epoch": 1.55,
"learning_rate": 1e-05,
"loss": 0.5022,
"step": 238
},
{
"epoch": 1.55,
"learning_rate": 9.929875738711565e-06,
"loss": 0.43,
"step": 239
},
{
"epoch": 1.56,
"learning_rate": 9.85975492576438e-06,
"loss": 0.5018,
"step": 240
},
{
"epoch": 1.56,
"learning_rate": 9.789641009330113e-06,
"loss": 0.3118,
"step": 241
},
{
"epoch": 1.57,
"learning_rate": 9.719537437241311e-06,
"loss": 0.301,
"step": 242
},
{
"epoch": 1.58,
"learning_rate": 9.64944765682184e-06,
"loss": 0.3905,
"step": 243
},
{
"epoch": 1.58,
"learning_rate": 9.579375114717352e-06,
"loss": 0.3628,
"step": 244
},
{
"epoch": 1.59,
"learning_rate": 9.50932325672582e-06,
"loss": 0.3875,
"step": 245
},
{
"epoch": 1.6,
"learning_rate": 9.439295527628083e-06,
"loss": 0.42,
"step": 246
},
{
"epoch": 1.6,
"learning_rate": 9.369295371018442e-06,
"loss": 0.2803,
"step": 247
},
{
"epoch": 1.61,
"learning_rate": 9.299326229135326e-06,
"loss": 0.3663,
"step": 248
},
{
"epoch": 1.62,
"learning_rate": 9.22939154269204e-06,
"loss": 0.3558,
"step": 249
},
{
"epoch": 1.62,
"learning_rate": 9.159494750707527e-06,
"loss": 0.4629,
"step": 250
},
{
"epoch": 1.63,
"learning_rate": 9.0896392903373e-06,
"loss": 0.2931,
"step": 251
},
{
"epoch": 1.64,
"learning_rate": 9.019828596704394e-06,
"loss": 0.4618,
"step": 252
},
{
"epoch": 1.64,
"learning_rate": 8.950066102730456e-06,
"loss": 0.4418,
"step": 253
},
{
"epoch": 1.65,
"learning_rate": 8.880355238966923e-06,
"loss": 0.4217,
"step": 254
},
{
"epoch": 1.66,
"learning_rate": 8.81069943342633e-06,
"loss": 0.3553,
"step": 255
},
{
"epoch": 1.66,
"learning_rate": 8.741102111413749e-06,
"loss": 0.3842,
"step": 256
},
{
"epoch": 1.67,
"learning_rate": 8.671566695358324e-06,
"loss": 0.4009,
"step": 257
},
{
"epoch": 1.68,
"learning_rate": 8.602096604645009e-06,
"loss": 0.3709,
"step": 258
},
{
"epoch": 1.68,
"learning_rate": 8.532695255446384e-06,
"loss": 0.3039,
"step": 259
},
{
"epoch": 1.69,
"learning_rate": 8.463366060554698e-06,
"loss": 0.2142,
"step": 260
},
{
"epoch": 1.69,
"learning_rate": 8.394112429214032e-06,
"loss": 0.5348,
"step": 261
},
{
"epoch": 1.7,
"learning_rate": 8.324937766952638e-06,
"loss": 0.4558,
"step": 262
},
{
"epoch": 1.71,
"learning_rate": 8.255845475415493e-06,
"loss": 0.3542,
"step": 263
},
{
"epoch": 1.71,
"learning_rate": 8.186838952197019e-06,
"loss": 0.3586,
"step": 264
},
{
"epoch": 1.72,
"learning_rate": 8.117921590674002e-06,
"loss": 0.3804,
"step": 265
},
{
"epoch": 1.73,
"learning_rate": 8.04909677983872e-06,
"loss": 0.3815,
"step": 266
},
{
"epoch": 1.73,
"learning_rate": 7.980367904132303e-06,
"loss": 0.3939,
"step": 267
},
{
"epoch": 1.74,
"learning_rate": 7.911738343278303e-06,
"loss": 0.5857,
"step": 268
},
{
"epoch": 1.75,
"learning_rate": 7.843211472116476e-06,
"loss": 0.2906,
"step": 269
},
{
"epoch": 1.75,
"learning_rate": 7.774790660436857e-06,
"loss": 0.3541,
"step": 270
},
{
"epoch": 1.76,
"learning_rate": 7.706479272814024e-06,
"loss": 0.3511,
"step": 271
},
{
"epoch": 1.77,
"learning_rate": 7.63828066844166e-06,
"loss": 0.3202,
"step": 272
},
{
"epoch": 1.77,
"learning_rate": 7.570198200967363e-06,
"loss": 0.2879,
"step": 273
},
{
"epoch": 1.78,
"learning_rate": 7.50223521832773e-06,
"loss": 0.3326,
"step": 274
},
{
"epoch": 1.79,
"learning_rate": 7.434395062583735e-06,
"loss": 0.353,
"step": 275
},
{
"epoch": 1.79,
"learning_rate": 7.366681069756352e-06,
"loss": 0.3996,
"step": 276
},
{
"epoch": 1.8,
"learning_rate": 7.299096569662549e-06,
"loss": 0.3469,
"step": 277
},
{
"epoch": 1.81,
"learning_rate": 7.2316448857515076e-06,
"loss": 0.3734,
"step": 278
},
{
"epoch": 1.81,
"learning_rate": 7.164329334941225e-06,
"loss": 0.3263,
"step": 279
},
{
"epoch": 1.82,
"learning_rate": 7.097153227455379e-06,
"loss": 0.2845,
"step": 280
},
{
"epoch": 1.82,
"learning_rate": 7.030119866660565e-06,
"loss": 0.3778,
"step": 281
},
{
"epoch": 1.83,
"learning_rate": 6.963232548903853e-06,
"loss": 0.2835,
"step": 282
},
{
"epoch": 1.84,
"learning_rate": 6.896494563350681e-06,
"loss": 0.3878,
"step": 283
},
{
"epoch": 1.84,
"learning_rate": 6.829909191823121e-06,
"loss": 0.489,
"step": 284
},
{
"epoch": 1.85,
"learning_rate": 6.763479708638485e-06,
"loss": 0.4006,
"step": 285
},
{
"epoch": 1.86,
"learning_rate": 6.697209380448333e-06,
"loss": 0.3494,
"step": 286
},
{
"epoch": 1.86,
"learning_rate": 6.631101466077801e-06,
"loss": 0.4427,
"step": 287
},
{
"epoch": 1.87,
"learning_rate": 6.5651592163653885e-06,
"loss": 0.4206,
"step": 288
},
{
"epoch": 1.88,
"learning_rate": 6.499385874003077e-06,
"loss": 0.4408,
"step": 289
},
{
"epoch": 1.88,
"learning_rate": 6.43378467337687e-06,
"loss": 0.2825,
"step": 290
},
{
"epoch": 1.89,
"learning_rate": 6.368358840407754e-06,
"loss": 0.4344,
"step": 291
},
{
"epoch": 1.9,
"learning_rate": 6.303111592393051e-06,
"loss": 0.2522,
"step": 292
},
{
"epoch": 1.9,
"learning_rate": 6.23804613784823e-06,
"loss": 0.351,
"step": 293
},
{
"epoch": 1.91,
"learning_rate": 6.173165676349103e-06,
"loss": 0.3388,
"step": 294
},
{
"epoch": 1.92,
"learning_rate": 6.108473398374509e-06,
"loss": 0.422,
"step": 295
},
{
"epoch": 1.92,
"learning_rate": 6.043972485149414e-06,
"loss": 0.4671,
"step": 296
},
{
"epoch": 1.93,
"learning_rate": 5.979666108488464e-06,
"loss": 0.4178,
"step": 297
},
{
"epoch": 1.94,
"learning_rate": 5.91555743064004e-06,
"loss": 0.4808,
"step": 298
},
{
"epoch": 1.94,
"learning_rate": 5.851649604130723e-06,
"loss": 0.31,
"step": 299
},
{
"epoch": 1.95,
"learning_rate": 5.787945771610296e-06,
"loss": 0.3015,
"step": 300
},
{
"epoch": 1.95,
"learning_rate": 5.724449065697182e-06,
"loss": 0.3486,
"step": 301
},
{
"epoch": 1.96,
"learning_rate": 5.66116260882442e-06,
"loss": 0.2433,
"step": 302
},
{
"epoch": 1.97,
"learning_rate": 5.598089513086108e-06,
"loss": 0.4164,
"step": 303
},
{
"epoch": 1.97,
"learning_rate": 5.535232880084373e-06,
"loss": 0.2996,
"step": 304
},
{
"epoch": 1.98,
"learning_rate": 5.472595800776831e-06,
"loss": 0.3965,
"step": 305
},
{
"epoch": 1.99,
"learning_rate": 5.410181355324622e-06,
"loss": 0.4008,
"step": 306
},
{
"epoch": 1.99,
"learning_rate": 5.347992612940927e-06,
"loss": 0.234,
"step": 307
},
{
"epoch": 2.0,
"learning_rate": 5.286032631740023e-06,
"loss": 0.34,
"step": 308
},
{
"epoch": 2.01,
"learning_rate": 5.22430445858694e-06,
"loss": 0.1256,
"step": 309
},
{
"epoch": 2.01,
"learning_rate": 5.1628111289476025e-06,
"loss": 0.1268,
"step": 310
},
{
"epoch": 2.02,
"learning_rate": 5.101555666739563e-06,
"loss": 0.1549,
"step": 311
},
{
"epoch": 2.03,
"learning_rate": 5.040541084183326e-06,
"loss": 0.144,
"step": 312
},
{
"epoch": 2.03,
"learning_rate": 4.979770381654181e-06,
"loss": 0.1679,
"step": 313
},
{
"epoch": 2.04,
"learning_rate": 4.919246547534709e-06,
"loss": 0.1874,
"step": 314
},
{
"epoch": 2.05,
"learning_rate": 4.858972558067784e-06,
"loss": 0.1988,
"step": 315
},
{
"epoch": 2.05,
"learning_rate": 4.798951377210253e-06,
"loss": 0.1511,
"step": 316
},
{
"epoch": 2.06,
"learning_rate": 4.739185956487169e-06,
"loss": 0.1088,
"step": 317
},
{
"epoch": 2.06,
"learning_rate": 4.679679234846636e-06,
"loss": 0.0947,
"step": 318
},
{
"epoch": 2.07,
"learning_rate": 4.6204341385153186e-06,
"loss": 0.1095,
"step": 319
},
{
"epoch": 2.08,
"learning_rate": 4.561453580854516e-06,
"loss": 0.1702,
"step": 320
},
{
"epoch": 2.08,
"learning_rate": 4.502740462216919e-06,
"loss": 0.1463,
"step": 321
},
{
"epoch": 2.09,
"learning_rate": 4.444297669803981e-06,
"loss": 0.1017,
"step": 322
},
{
"epoch": 2.1,
"learning_rate": 4.386128077523923e-06,
"loss": 0.1365,
"step": 323
},
{
"epoch": 2.1,
"learning_rate": 4.328234545850441e-06,
"loss": 0.0989,
"step": 324
},
{
"epoch": 2.11,
"learning_rate": 4.270619921682019e-06,
"loss": 0.138,
"step": 325
},
{
"epoch": 2.12,
"learning_rate": 4.213287038201943e-06,
"loss": 0.1333,
"step": 326
},
{
"epoch": 2.12,
"learning_rate": 4.156238714738974e-06,
"loss": 0.1526,
"step": 327
},
{
"epoch": 2.13,
"learning_rate": 4.09947775662872e-06,
"loss": 0.1285,
"step": 328
},
{
"epoch": 2.14,
"learning_rate": 4.043006955075667e-06,
"loss": 0.1414,
"step": 329
},
{
"epoch": 2.14,
"learning_rate": 3.986829087015941e-06,
"loss": 0.1694,
"step": 330
},
{
"epoch": 2.15,
"learning_rate": 3.930946914980744e-06,
"loss": 0.1222,
"step": 331
},
{
"epoch": 2.16,
"learning_rate": 3.875363186960499e-06,
"loss": 0.1736,
"step": 332
},
{
"epoch": 2.16,
"learning_rate": 3.820080636269737e-06,
"loss": 0.1222,
"step": 333
},
{
"epoch": 2.17,
"learning_rate": 3.7651019814126656e-06,
"loss": 0.1431,
"step": 334
},
{
"epoch": 2.18,
"learning_rate": 3.7104299259495113e-06,
"loss": 0.1902,
"step": 335
},
{
"epoch": 2.18,
"learning_rate": 3.6560671583635467e-06,
"loss": 0.1266,
"step": 336
},
{
"epoch": 2.19,
"learning_rate": 3.6020163519289077e-06,
"loss": 0.1607,
"step": 337
},
{
"epoch": 2.19,
"learning_rate": 3.5482801645791266e-06,
"loss": 0.1206,
"step": 338
},
{
"epoch": 2.2,
"learning_rate": 3.494861238776418e-06,
"loss": 0.1484,
"step": 339
},
{
"epoch": 2.21,
"learning_rate": 3.4417622013817597e-06,
"loss": 0.1453,
"step": 340
},
{
"epoch": 2.21,
"learning_rate": 3.3889856635257024e-06,
"loss": 0.1299,
"step": 341
},
{
"epoch": 2.22,
"learning_rate": 3.3365342204799613e-06,
"loss": 0.1442,
"step": 342
},
{
"epoch": 2.23,
"learning_rate": 3.284410451529816e-06,
"loss": 0.1326,
"step": 343
},
{
"epoch": 2.23,
"learning_rate": 3.2326169198472555e-06,
"loss": 0.1413,
"step": 344
},
{
"epoch": 2.24,
"learning_rate": 3.1811561723649496e-06,
"loss": 0.1679,
"step": 345
},
{
"epoch": 2.25,
"learning_rate": 3.1300307396509833e-06,
"loss": 0.1357,
"step": 346
},
{
"epoch": 2.25,
"learning_rate": 3.0792431357844444e-06,
"loss": 0.1545,
"step": 347
},
{
"epoch": 2.26,
"learning_rate": 3.028795858231768e-06,
"loss": 0.1363,
"step": 348
},
{
"epoch": 2.27,
"learning_rate": 2.9786913877239486e-06,
"loss": 0.1149,
"step": 349
},
{
"epoch": 2.27,
"learning_rate": 2.9289321881345257e-06,
"loss": 0.2117,
"step": 350
},
{
"epoch": 2.28,
"learning_rate": 2.879520706358446e-06,
"loss": 0.0599,
"step": 351
},
{
"epoch": 2.29,
"learning_rate": 2.8304593721917283e-06,
"loss": 0.0972,
"step": 352
},
{
"epoch": 2.29,
"learning_rate": 2.7817505982119708e-06,
"loss": 0.1344,
"step": 353
},
{
"epoch": 2.3,
"learning_rate": 2.7333967796597317e-06,
"loss": 0.1688,
"step": 354
},
{
"epoch": 2.31,
"learning_rate": 2.6854002943207245e-06,
"loss": 0.221,
"step": 355
},
{
"epoch": 2.31,
"learning_rate": 2.637763502408909e-06,
"loss": 0.1754,
"step": 356
},
{
"epoch": 2.32,
"learning_rate": 2.5904887464504115e-06,
"loss": 0.1275,
"step": 357
},
{
"epoch": 2.32,
"learning_rate": 2.5435783511683444e-06,
"loss": 0.1477,
"step": 358
},
{
"epoch": 2.33,
"learning_rate": 2.4970346233684863e-06,
"loss": 0.1861,
"step": 359
},
{
"epoch": 2.34,
"learning_rate": 2.450859851825842e-06,
"loss": 0.1429,
"step": 360
},
{
"epoch": 2.34,
"learning_rate": 2.4050563071720867e-06,
"loss": 0.1739,
"step": 361
},
{
"epoch": 2.35,
"learning_rate": 2.3596262417839256e-06,
"loss": 0.1637,
"step": 362
},
{
"epoch": 2.36,
"learning_rate": 2.3145718896723204e-06,
"loss": 0.1599,
"step": 363
},
{
"epoch": 2.36,
"learning_rate": 2.26989546637263e-06,
"loss": 0.1504,
"step": 364
},
{
"epoch": 2.37,
"learning_rate": 2.225599168835677e-06,
"loss": 0.1674,
"step": 365
},
{
"epoch": 2.38,
"learning_rate": 2.1816851753197023e-06,
"loss": 0.1095,
"step": 366
},
{
"epoch": 2.38,
"learning_rate": 2.138155645283244e-06,
"loss": 0.1598,
"step": 367
},
{
"epoch": 2.39,
"learning_rate": 2.095012719278966e-06,
"loss": 0.1087,
"step": 368
},
{
"epoch": 2.4,
"learning_rate": 2.0522585188483745e-06,
"loss": 0.1596,
"step": 369
},
{
"epoch": 2.4,
"learning_rate": 2.009895146417512e-06,
"loss": 0.1298,
"step": 370
},
{
"epoch": 2.41,
"learning_rate": 1.967924685193552e-06,
"loss": 0.1419,
"step": 371
},
{
"epoch": 2.42,
"learning_rate": 1.9263491990623763e-06,
"loss": 0.1005,
"step": 372
},
{
"epoch": 2.42,
"learning_rate": 1.885170732487074e-06,
"loss": 0.132,
"step": 373
},
{
"epoch": 2.43,
"learning_rate": 1.8443913104073984e-06,
"loss": 0.1426,
"step": 374
},
{
"epoch": 2.44,
"learning_rate": 1.8040129381402137e-06,
"loss": 0.1606,
"step": 375
},
{
"epoch": 2.44,
"learning_rate": 1.7640376012808536e-06,
"loss": 0.1443,
"step": 376
},
{
"epoch": 2.45,
"learning_rate": 1.7244672656055105e-06,
"loss": 0.1098,
"step": 377
},
{
"epoch": 2.45,
"learning_rate": 1.6853038769745466e-06,
"loss": 0.1869,
"step": 378
},
{
"epoch": 2.46,
"learning_rate": 1.6465493612368233e-06,
"loss": 0.1249,
"step": 379
},
{
"epoch": 2.47,
"learning_rate": 1.6082056241349787e-06,
"loss": 0.1609,
"step": 380
},
{
"epoch": 2.47,
"learning_rate": 1.5702745512117323e-06,
"loss": 0.1411,
"step": 381
},
{
"epoch": 2.48,
"learning_rate": 1.5327580077171589e-06,
"loss": 0.1534,
"step": 382
},
{
"epoch": 2.49,
"learning_rate": 1.495657838516953e-06,
"loss": 0.1766,
"step": 383
},
{
"epoch": 2.49,
"learning_rate": 1.4589758680017263e-06,
"loss": 0.092,
"step": 384
},
{
"epoch": 2.5,
"learning_rate": 1.4227138999972801e-06,
"loss": 0.1525,
"step": 385
},
{
"epoch": 2.51,
"learning_rate": 1.3868737176759105e-06,
"loss": 0.0968,
"step": 386
},
{
"epoch": 2.51,
"learning_rate": 1.3514570834687203e-06,
"loss": 0.1209,
"step": 387
},
{
"epoch": 2.52,
"learning_rate": 1.3164657389789459e-06,
"loss": 0.1049,
"step": 388
},
{
"epoch": 2.53,
"learning_rate": 1.281901404896323e-06,
"loss": 0.137,
"step": 389
},
{
"epoch": 2.53,
"learning_rate": 1.2477657809124632e-06,
"loss": 0.0888,
"step": 390
},
{
"epoch": 2.54,
"learning_rate": 1.2140605456372856e-06,
"loss": 0.1431,
"step": 391
},
{
"epoch": 2.55,
"learning_rate": 1.1807873565164507e-06,
"loss": 0.1338,
"step": 392
},
{
"epoch": 2.55,
"learning_rate": 1.1479478497498796e-06,
"loss": 0.1126,
"step": 393
},
{
"epoch": 2.56,
"learning_rate": 1.1155436402112785e-06,
"loss": 0.1365,
"step": 394
},
{
"epoch": 2.56,
"learning_rate": 1.08357632136873e-06,
"loss": 0.1151,
"step": 395
},
{
"epoch": 2.57,
"learning_rate": 1.0520474652063395e-06,
"loss": 0.1358,
"step": 396
},
{
"epoch": 2.58,
"learning_rate": 1.0209586221469336e-06,
"loss": 0.1445,
"step": 397
},
{
"epoch": 2.58,
"learning_rate": 9.903113209758098e-07,
"loss": 0.1223,
"step": 398
},
{
"epoch": 2.59,
"learning_rate": 9.601070687655667e-07,
"loss": 0.0742,
"step": 399
},
{
"epoch": 2.6,
"learning_rate": 9.303473508019944e-07,
"loss": 0.0628,
"step": 400
},
{
"epoch": 2.6,
"learning_rate": 9.010336305110345e-07,
"loss": 0.1212,
"step": 401
},
{
"epoch": 2.61,
"learning_rate": 8.721673493868111e-07,
"loss": 0.1515,
"step": 402
},
{
"epoch": 2.62,
"learning_rate": 8.437499269207538e-07,
"loss": 0.1941,
"step": 403
},
{
"epoch": 2.62,
"learning_rate": 8.157827605317892e-07,
"loss": 0.1574,
"step": 404
},
{
"epoch": 2.63,
"learning_rate": 7.882672254976298e-07,
"loss": 0.0986,
"step": 405
},
{
"epoch": 2.64,
"learning_rate": 7.612046748871327e-07,
"loss": 0.1015,
"step": 406
},
{
"epoch": 2.64,
"learning_rate": 7.345964394937788e-07,
"loss": 0.1234,
"step": 407
},
{
"epoch": 2.65,
"learning_rate": 7.084438277702188e-07,
"loss": 0.129,
"step": 408
},
{
"epoch": 2.66,
"learning_rate": 6.827481257639346e-07,
"loss": 0.1382,
"step": 409
},
{
"epoch": 2.66,
"learning_rate": 6.57510597054003e-07,
"loss": 0.1416,
"step": 410
},
{
"epoch": 2.67,
"learning_rate": 6.327324826889469e-07,
"loss": 0.1398,
"step": 411
},
{
"epoch": 2.68,
"learning_rate": 6.084150011257239e-07,
"loss": 0.1144,
"step": 412
},
{
"epoch": 2.68,
"learning_rate": 5.845593481697931e-07,
"loss": 0.1226,
"step": 413
},
{
"epoch": 2.69,
"learning_rate": 5.611666969163243e-07,
"loss": 0.1322,
"step": 414
},
{
"epoch": 2.69,
"learning_rate": 5.382381976925044e-07,
"loss": 0.1468,
"step": 415
},
{
"epoch": 2.7,
"learning_rate": 5.157749780009735e-07,
"loss": 0.1106,
"step": 416
},
{
"epoch": 2.71,
"learning_rate": 4.937781424643728e-07,
"loss": 0.1264,
"step": 417
},
{
"epoch": 2.71,
"learning_rate": 4.7224877277103673e-07,
"loss": 0.1333,
"step": 418
},
{
"epoch": 2.72,
"learning_rate": 4.511879276217967e-07,
"loss": 0.1185,
"step": 419
},
{
"epoch": 2.73,
"learning_rate": 4.305966426779118e-07,
"loss": 0.1008,
"step": 420
},
{
"epoch": 2.73,
"learning_rate": 4.1047593051015245e-07,
"loss": 0.1221,
"step": 421
},
{
"epoch": 2.74,
"learning_rate": 3.908267805490051e-07,
"loss": 0.1004,
"step": 422
},
{
"epoch": 2.75,
"learning_rate": 3.7165015903600553e-07,
"loss": 0.1142,
"step": 423
},
{
"epoch": 2.75,
"learning_rate": 3.529470089762421e-07,
"loss": 0.1041,
"step": 424
},
{
"epoch": 2.76,
"learning_rate": 3.347182500919677e-07,
"loss": 0.0949,
"step": 425
},
{
"epoch": 2.77,
"learning_rate": 3.1696477877738664e-07,
"loss": 0.106,
"step": 426
},
{
"epoch": 2.77,
"learning_rate": 2.996874680545603e-07,
"loss": 0.1299,
"step": 427
},
{
"epoch": 2.78,
"learning_rate": 2.8288716753049007e-07,
"loss": 0.1054,
"step": 428
},
{
"epoch": 2.79,
"learning_rate": 2.665647033553309e-07,
"loss": 0.1618,
"step": 429
},
{
"epoch": 2.79,
"learning_rate": 2.507208781817638e-07,
"loss": 0.0967,
"step": 430
},
{
"epoch": 2.8,
"learning_rate": 2.3535647112553295e-07,
"loss": 0.1235,
"step": 431
},
{
"epoch": 2.81,
"learning_rate": 2.20472237727124e-07,
"loss": 0.1017,
"step": 432
},
{
"epoch": 2.81,
"learning_rate": 2.0606890991461737e-07,
"loss": 0.1592,
"step": 433
},
{
"epoch": 2.82,
"learning_rate": 1.921471959676957e-07,
"loss": 0.1495,
"step": 434
},
{
"epoch": 2.82,
"learning_rate": 1.787077804828097e-07,
"loss": 0.152,
"step": 435
},
{
"epoch": 2.83,
"learning_rate": 1.657513243395159e-07,
"loss": 0.1158,
"step": 436
},
{
"epoch": 2.84,
"learning_rate": 1.5327846466797857e-07,
"loss": 0.1455,
"step": 437
},
{
"epoch": 2.84,
"learning_rate": 1.4128981481764115e-07,
"loss": 0.1579,
"step": 438
},
{
"epoch": 2.85,
"learning_rate": 1.2978596432705826e-07,
"loss": 0.1225,
"step": 439
},
{
"epoch": 2.86,
"learning_rate": 1.1876747889491225e-07,
"loss": 0.1159,
"step": 440
},
{
"epoch": 2.86,
"learning_rate": 1.0823490035218986e-07,
"loss": 0.1001,
"step": 441
},
{
"epoch": 2.87,
"learning_rate": 9.818874663554356e-08,
"loss": 0.106,
"step": 442
},
{
"epoch": 2.88,
"learning_rate": 8.862951176181744e-08,
"loss": 0.089,
"step": 443
},
{
"epoch": 2.88,
"learning_rate": 7.955766580375334e-08,
"loss": 0.0875,
"step": 444
},
{
"epoch": 2.89,
"learning_rate": 7.097365486688158e-08,
"loss": 0.1156,
"step": 445
},
{
"epoch": 2.9,
"learning_rate": 6.287790106757396e-08,
"loss": 0.1663,
"step": 446
},
{
"epoch": 2.9,
"learning_rate": 5.527080251229833e-08,
"loss": 0.1311,
"step": 447
},
{
"epoch": 2.91,
"learning_rate": 4.815273327803183e-08,
"loss": 0.1318,
"step": 448
},
{
"epoch": 2.92,
"learning_rate": 4.152404339386795e-08,
"loss": 0.1204,
"step": 449
},
{
"epoch": 2.92,
"learning_rate": 3.538505882380916e-08,
"loss": 0.1286,
"step": 450
},
{
"epoch": 2.93,
"learning_rate": 2.9736081450730813e-08,
"loss": 0.1725,
"step": 451
},
{
"epoch": 2.94,
"learning_rate": 2.4577389061539724e-08,
"loss": 0.1702,
"step": 452
},
{
"epoch": 2.94,
"learning_rate": 1.9909235333517296e-08,
"loss": 0.1728,
"step": 453
},
{
"epoch": 2.95,
"learning_rate": 1.5731849821833955e-08,
"loss": 0.1095,
"step": 454
},
{
"epoch": 2.95,
"learning_rate": 1.2045437948275952e-08,
"loss": 0.1152,
"step": 455
},
{
"epoch": 2.96,
"learning_rate": 8.850180991131219e-09,
"loss": 0.1073,
"step": 456
},
{
"epoch": 2.97,
"learning_rate": 6.146236076279843e-09,
"loss": 0.1369,
"step": 457
},
{
"epoch": 2.97,
"learning_rate": 3.933736169471347e-09,
"loss": 0.1724,
"step": 458
},
{
"epoch": 2.98,
"learning_rate": 2.2127900697777038e-09,
"loss": 0.1503,
"step": 459
},
{
"epoch": 2.99,
"learning_rate": 9.83482404249836e-10,
"loss": 0.1022,
"step": 460
},
{
"epoch": 2.99,
"learning_rate": 2.45873623754278e-10,
"loss": 0.1463,
"step": 461
},
{
"epoch": 3.0,
"learning_rate": 0.0,
"loss": 0.0895,
"step": 462
},
{
"epoch": 3.0,
"step": 462,
"total_flos": 2.1534105216745472e+17,
"train_loss": 0.6429303579548479,
"train_runtime": 2956.9791,
"train_samples_per_second": 19.94,
"train_steps_per_second": 0.156
}
],
"logging_steps": 1.0,
"max_steps": 462,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 100,
"total_flos": 2.1534105216745472e+17,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}